Fig. 1. Box plot and principal component analysis (PCA) using normalized fragments per kilobase of transcript per million mapped fragments (FPKM) before (A) and after (B) batch correction in 17 samples. Samples with SRP197370, SRP083865 and SRP117202 were from Zeng et al (2019b), Nanda et al (2017) and Lu et al (2021), respectively. Only samples in the first two PCA axis were shown.

Fig. 2. miRNAs targeting the identified conserved differentially expressed modular genes. miRNAs were retrieved from http://structuralbiology.cau.edu.cn/PlantGSEA/database (Yi et al, 2013).

Fig. 3. qRT-PCR validation of RNA-Seq expression of conserved differentially expressed modular genes. RNA-Seq fold changes, which have been reported for root (Table S5) and crown (Table S7) in Nanda et al (2017) and Zeng et al (2019b), respectively, were compared with the obtained qRT-PCR based fold change for the same tissues. During qRT-PCR, expression levels under the control conditions (Zn supply) were normalized to 1 for both root and crown tissues. ACTN-1 was used as an internal reference. The Pearson correlation indicated a positive significant (P < 0.05) correlation (R2 = 0.87) between fold changes obtained based on RNA-Seq and qRT-PCR analyses.

Fig. 4. A theoretical model for signaling pathways underlying Zn deficiency responses in rice. Zn deficiency stress first activates regulatory proteins, such as miRNAs, protein kinases, and transcription factors that modulate various hormonal stress responses, sugar transports, and morphogenetic events. Following that, regulatory proteins induce the expression of the downstream Zn responsive genes, including the small-molecule metabolic process genes, Zn transporter genes, transmembrane transporter genes, metal binding and reactive oxygen species scavenging genes, ion or chemical transporter genes and sugar transporter genes.

Fig. 5. An illustration of potential use of conserved modular differentially expressed genes (DEGs) in breeding schemes for high grain Zn. Co-expression network analysis allows identification of modules of co-expressed genes. By intersecting modular genes with DEGs reported in different studies on Zn deficiency stress, conserved modular DEGs are obtained. In the next stage, grain Zn-relevant markers can be obtained by selecting features in the proximities of conserved modular DEGs. Finally, the functional markers identified can be deployed in genomic selection models to improve prediction accuracy of high grain Zn lines. Besides genomic selection-based methods, if functionally validated, conserved modular DEGs could also facilitate the development of high-grain Zn via genome editing or transgenic methods. However, food regulations in most developing countries are likely to obstruct delivery of genome-edited or genetically modified biofortified crops.